Singal B M, Hedges J R, Radack K L
Department of Emergency Medicine, University of Cincinnati, College of Medicine, Ohio 45267-0769.
Ann Emerg Med. 1989 Jan;18(1):13-20. doi: 10.1016/s0196-0644(89)80304-x.
Prediction rules have been recommended for guiding the ordering of diagnostic tests. Such rules can be used to define low-yield criteria (LYC) for the purpose of identifying patients with an extremely low probability of disease and hence discouraging test ordering by the physician on patients meeting LYC. In this study, community hospital emergency department populations of adults (n = 255) and children (n = 78) were evaluated prospectively for the presence of predictive clinical parameters and the physician's estimate of pneumonia prior to obtaining a chest film. We developed LYC and analyzed published LYC for obtaining chest films on patients considered at risk for pneumonia by means of logistic regression, receiver operating characteristics curve, and negative predictive value analyses. We were unable to derive or validate clinically useful LYC to improve on the seasoned clinician's probability estimate of pneumonia. We discuss the inherent limitations in the development and application of LYC that must be understood by those who seek to limit the ordering of chest films by the application of guidelines developed from decision rules.
已有预测规则被推荐用于指导诊断检查的安排。此类规则可用于定义低收益标准（LYC），以识别疾病概率极低的患者，从而劝阻医生对符合LYC的患者进行检查。在本研究中，对社区医院急诊科的成年人群（n = 255）和儿童人群（n = 78）进行了前瞻性评估，以确定在获取胸部X光片之前是否存在预测性临床参数以及医生对肺炎的判断。我们制定了LYC，并通过逻辑回归、受试者操作特征曲线和阴性预测值分析，分析了已发表的关于对有肺炎风险患者进行胸部X光检查的LYC。我们无法得出或验证临床上有用的LYC来改进经验丰富的临床医生对肺炎的概率判断。我们讨论了LYC开发和应用中固有的局限性，那些试图通过应用基于决策规则制定的指南来限制胸部X光检查安排的人必须了解这些局限性。
